On Tue, Jun 21, 2005 at 09:24:35PM -0400, Jacques Mony wrote:
So there hasn't been development for a while now. And, the
discussion on the
future of uuu was kind of short.
What I suggest is we go back to the roots. A JiT modular OS that
will have
a JiT engine for each platform it runs on. I would remove all
posix-compliant stuff. Let's do it from scratch for real, so we
don't need
to bother about interfaces of any kind. Yes, developing driver
will take
long, but as long as we have QEMU support, we can go far.
The current plan isn't to make Uuu posix compliant; it is to make
it run
*on* posix. This makes things easier because all the boring problems
that have been solved a million times like drivers and memory
management
and booting don't need to be solved again. It makes debugging
easier and
makes the compile-test-debug cycle about 50 times faster.
Likewise, the JIT problem has already been solved. mono runs on plenty
of platforms, performs pretty well, and mostly runs python through
ironpython. Getting it to run in unununium isn't too hard; the
difficulty lies in the number of libraries on which it depends which
must also be tinkered to work.
Not using these solutions other people have provided does not mean
there
are no interfaces to bother us. It just means that the interfaces are
poorly documented, often propriatary, extremely low level, and vary
greatly from individual machine to machine.
The way to get unununium to develop more is to write code. The case
now
is not that there are difficulties with the current design; it's
that no
one is even trying to develop it. Most of the more recent design
decisions were made with the goal of getting unununium done sooner.
The problem with this, if any, is that in doing so unununium has lost
its appeal to people who want to implement an entire program from the
lowest level for the sake of tinkering. I was once one of those
people,
but I have sense learned my lessons. I have written scores of
bootloaders, memory managers, schedulers, disk drivers, and filesystem
implementations. I know how all that stuff works now, and I don't want
to do it anymore. I want to explore new ideas.
The reason unununium isn't moving currently is because I'm busy
exploring new ideas in unrelated projects. Anyone is welcome to
pick up
the project, and I am quite willing to answer any good questions about
the existing code or design. If that doesn't happen, unununium will
just
have to wait until my interest falls on it again either by chance or
piles of cash.
_______________________________________________
Uuu-devel mailing list
Uuu-devel@xxxxxxxxxxxxx
http://unununium.org/cgi-bin/mailman/listinfo/uuu-devel
